PyTorchKR Papers

최신 논문

총 5,576편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2108.08607

Generating Superpixels for High-resolution Images with Decoupled Patch Calibration

Superpixel segmentation has recently seen important progress benefiting from the advances in differentiable deep learning. However, the very high-resolution superpixel segmentation still remains challenging due to the expensive memory and computation cost, making the current advanced superpixel networks fail to process. In this paper, we devise Patch Calibration Networks (PCNet), aiming to efficiently and accurately implement high-resolution superpixel segmentation. PCNet follows the principle of producing high-resolution output from low-resolution input for saving GPU memory and relieving computation cost. To recall the fine details destroyed by the down-sampling operation, we propose a novel Decoupled Patch Calibration (DPC) branch for collaboratively augment the main superpixel generation branch. In particular, DPC takes a local patch from the high-resolution images and dynamically generates a binary mask to impose the network to focus on region boundaries. By sharing the parameters of DPC and main branches, the fine-detailed knowledge learned from high-resolution patches will be transferred to help calibrate the destroyed information. To the best of our knowledge, we make the first attempt to consider the deep-learning-based superpixel generation for high-resolution cases. To facilitate this research, we build evaluation benchmarks from two public datasets and one new constructed one, covering a wide range of diversities from fine-grained human parts to cityscapes. Extensive experiments demonstrate that our PCNet can not only perform favorably against the state-of-the-arts in the quantitative results but also improve the resolution upper bound from 3K to 5K on 1080Ti GPUs.

arXiv:2108.07004

Real-time Transmission of Geometrically-shaped Signals using a Software-defined GPU-based Optical Receiver

A software-defined optical receiver is implemented on an off-the-shelf commercial graphics processing unit (GPU). The receiver provides real-time signal processing functionality to process 1 GBaud minimum phase (MP) 4-, 8-, 16-, 32-, 64-, 128-ary quadrature amplitude modulation (QAM) as well as geometrically shaped (GS) 8- and 128-QAM signals using Kramers-Kronig (KK) coherent detection. Experimental validation of this receiver over a 91~km field-deployed optical fiber link between two Tokyo locations is shown with detailed optical signal-to-noise ratio (OSNR) investigations. A net data rate of 5 Gbps using 64-QAM is demonstrated.

arXiv:2108.07001

Real-time 10,000 km Straight-line Transmission using a Software-defined GPU-Based Receiver

Real-time 10,000 km transmission over a straight-line link is achieved using a software-defined multi-modulation format receiver implemented on a commercial off-the-shelf general-purpose graphics processing unit (GPU). Minimum phase 1 GBaud 4-ary quadrature amplitude modulation (QAM) signals are transmitted over 10,000 km and successfully received after detection with a Kramers-Kronig (KK) coherent receiver. 8-, 16-, 32-, and 64-QAM are successfully transmitted over 7600, 5600, 3600, and 1600 km, respectively.

arXiv:2108.06610공식 코드 · Papers with Code

SquiggleFilter: An Accelerator for Portable Virus Detection

The MinION is a recent-to-market handheld nanopore sequencer. It can be used to determine the whole genome of a target virus in a biological sample. Its Read Until feature allows us to skip sequencing a majority of non-target reads (DNA/RNA fragments), which constitutes more than 99% of all reads in a typical sample. However, it does not have any on-board computing, which significantly limits its portability. We analyze the performance of a Read Until metagenomic pipeline for detecting target viruses and identifying strain-specific mutations. We find new sources of performance bottlenecks (basecaller in classification of a read) that are not addressed by past genomics accelerators. We present SquiggleFilter, a novel hardware accelerated dynamic time warping (DTW) based filter that directly analyzes MinION's raw squiggles and filters everything except target viral reads, thereby avoiding the expensive basecalling step. We show that our 14.3W 13.25mm2 accelerator has 274X greater throughput and 3481X lower latency than existing GPU-based solutions while consuming half the power, enabling Read Until for the next generation of nanopore sequencers.

arXiv:2108.06455공식 코드 · Papers with Code

PTT: Point-Track-Transformer Module for 3D Single Object Tracking in Point Clouds

3D single object tracking is a key issue for robotics. In this paper, we propose a transformer module called Point-Track-Transformer (PTT) for point cloud-based 3D single object tracking. PTT module contains three blocks for feature embedding, position encoding, and self-attention feature computation. Feature embedding aims to place features closer in the embedding space if they have similar semantic information. Position encoding is used to encode coordinates of point clouds into high dimension distinguishable features. Self-attention generates refined attention features by computing attention weights. Besides, we embed the PTT module into the open-source state-of-the-art method P2B to construct PTT-Net. Experiments on the KITTI dataset reveal that our PTT-Net surpasses the state-of-the-art by a noticeable margin (~10%). Additionally, PTT-Net could achieve real-time performance (~40FPS) on NVIDIA 1080Ti GPU. Our code is open-sourced for the robotics community at https://github.com/shanjiayao/PTT.

arXiv:2108.06156공식 코드 · Papers with Code

EEEA-Net: An Early Exit Evolutionary Neural Architecture Search

The goals of this research were to search for Convolutional Neural Network (CNN) architectures, suitable for an on-device processor with limited computing resources, performing at substantially lower Network Architecture Search (NAS) costs. A new algorithm entitled an Early Exit Population Initialisation (EE-PI) for Evolutionary Algorithm (EA) was developed to achieve both goals. The EE-PI reduces the total number of parameters in the search process by filtering the models with fewer parameters than the maximum threshold. It will look for a new model to replace those models with parameters more than the threshold. Thereby, reducing the number of parameters, memory usage for model storage and processing time while maintaining the same performance or accuracy. The search time was reduced to 0.52 GPU day. This is a huge and significant achievement compared to the NAS of 4 GPU days achieved using NSGA-Net, 3,150 GPU days by the AmoebaNet model, and the 2,000 GPU days by the NASNet model. As well, Early Exit Evolutionary Algorithm networks (EEEA-Nets) yield network architectures with minimal error and computational cost suitable for a given dataset as a class of network algorithms. Using EEEA-Net on CIFAR-10, CIFAR-100, and ImageNet datasets, our experiments showed that EEEA-Net achieved the lowest error rate among state-of-the-art NAS models, with 2.46% for CIFAR-10, 15.02% for CIFAR-100, and 23.8% for ImageNet dataset. Further, we implemented this image recognition architecture for other tasks, such as object detection, semantic segmentation, and keypoint detection tasks, and, in our experiments, EEEA-Net-C2 outperformed MobileNet-V3 on all of these various tasks. (The algorithm code is available at https://github.com/chakkritte/EEEA-Net).

arXiv:2108.05818공식 코드 · Papers with Code

PatrickStar: Parallel Training of Pre-trained Models via Chunk-based Memory Management

The pre-trained model (PTM) is revolutionizing Artificial Intelligence (AI) technology. However, the hardware requirement of PTM training is prohibitively high, making it a game for a small proportion of people. Therefore, we proposed PatrickStar system to lower the hardware requirements of PTMs and make them accessible to everyone. PatrickStar uses the CPU-GPU heterogeneous memory space to store the model data. Different from existing works, we organize the model data in memory chunks and dynamically distribute them in the heterogeneous memory. Guided by the runtime memory statistics collected in a warm-up iteration, chunks are orchestrated efficiently in heterogeneous memory and generate lower CPU-GPU data transmission volume and higher bandwidth utilization. Symbiosis with the Zero Redundancy Optimizer, PatrickStar scales to multiple GPUs on multiple nodes. % using data parallelism. The system can train tasks on bigger models and larger batch sizes, which cannot be accomplished by existing works. Experimental results show that PatrickStar extends model scales 2.27 and 2.5 times of DeepSpeed, and consistently exhibits significantly higher execution speed. PatricStar also successfully runs the 175B GPT3 training task on a 32 GPU cluster. Our code is publicly available at https://github.com/Tencent/PatrickStar.

arXiv:2108.05269공식 코드 · Papers with Code

Learning to Rearrange Voxels in Binary Segmentation Masks for Smooth Manifold Triangulation

Medical images, especially volumetric images, are of high resolution and often exceed the capacity of standard desktop GPUs. As a result, most deep learning-based medical image analysis tasks require the input images to be downsampled, often substantially, before these can be fed to a neural network. However, downsampling can lead to a loss of image quality, which is undesirable especially in reconstruction tasks, where the fine geometric details need to be preserved. In this paper, we propose that high-resolution images can be reconstructed in a coarse-to-fine fashion, where a deep learning algorithm is only responsible for generating a coarse representation of the image, which consumes moderate GPU memory. For producing the high-resolution outcome, we propose two novel methods: learned voxel rearrangement of the coarse output and hierarchical image synthesis. Compared to the coarse output, the high-resolution counterpart allows for smooth surface triangulation, which can be 3D-printed in the highest possible quality. Experiments of this paper are carried out on the dataset of AutoImplant 2021 (https://autoimplant2021.grand-challenge.org/), a MICCAI challenge on cranial implant design. The dataset contains high-resolution skulls that can be viewed as 2D manifolds embedded in a 3D space. Codes associated with this study can be accessed at https://github.com/Jianningli/voxel_rearrangement.

arXiv:2108.05170

ProAI: An Efficient Embedded AI Hardware for Automotive Applications -- a Benchmark Study

Development in the field of Single Board Computers (SBC) have been increasing for several years. They provide a good balance between computing performance and power consumption which is usually required for mobile platforms, like application in vehicles for Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD). However, there is an ever-increasing need of more powerful and efficient SBCs which can run power intensive Deep Neural Networks (DNNs) in real-time and can also satisfy necessary functional safety requirements such as Automotive Safety Integrity Level (ASIL). ProAI is being developed by ZF mainly to run powerful and efficient applications such as multitask DNNs and on top of that it also has the required safety certification for AD. In this work, we compare and discuss state of the art SBC on the basis of power intensive multitask DNN architecture called Multitask-CenterNet with respect to performance measures such as, FPS and power efficiency. As an automotive supercomputer, ProAI delivers an excellent combination of performance and efficiency, managing nearly twice the number of FPS per watt than a modern workstation laptop and almost four times compared to the Jetson Nano. Furthermore, it was also shown that there is still power in reserve for further and more complex tasks on the ProAI, based on the CPU and GPU utilization during the benchmark.

arXiv:2108.04541공식 코드 · Papers with Code

Accelerating Evolutionary Neural Architecture Search via Multi-Fidelity Evaluation

Evolutionary neural architecture search (ENAS) has recently received increasing attention by effectively finding high-quality neural architectures, which however consumes high computational cost by training the architecture encoded by each individual for complete epochs in individual evaluation. Numerous ENAS approaches have been developed to reduce the evaluation cost, but it is often difficult for most of these approaches to achieve high evaluation accuracy. To address this issue, in this paper we propose an accelerated ENAS via multifidelity evaluation termed MFENAS, where the individual evaluation cost is significantly reduced by training the architecture encoded by each individual for only a small number of epochs. The balance between evaluation cost and evaluation accuracy is well maintained by suggesting a multi-fidelity evaluation, which identifies the potentially good individuals that cannot survive from previous generations by integrating multiple evaluations under different numbers of training epochs. For high diversity of neural architectures, a population initialization strategy is devised to produce different neural architectures varying from ResNet-like architectures to Inception-like ones. Experimental results on CIFAR-10 show that the architecture obtained by the proposed MFENAS achieves a 2.39% test error rate at the cost of only 0.6 GPU days on one NVIDIA 2080TI GPU, demonstrating the superiority of the proposed MFENAS over state-of-the-art NAS approaches in terms of both computational cost and architecture quality. The architecture obtained by the proposed MFENAS is then transferred to CIFAR-100 and ImageNet, which also exhibits competitive performance to the architectures obtained by existing NAS approaches. The source code of the proposed MFENAS is available at https://github.com/DevilYangS/MFENAS/.

arXiv:2108.04315

FL-MISR: Fast Large-Scale Multi-Image Super-Resolution for Computed Tomography Based on Multi-GPU Acceleration

Multi-image super-resolution (MISR) usually outperforms single-image super-resolution (SISR) under a proper inter-image alignment by explicitly exploiting the inter-image correlation. However, the large computational demand encumbers the deployment of MISR in practice. In this work, we propose a distributed optimization framework based on data parallelism for fast large-scale MISR using multi-GPU acceleration named FL-MISR. The scaled conjugate gradient (SCG) algorithm is applied to the distributed subfunctions and the local SCG variables are communicated to synchronize the convergence rate over multi-GPU systems towards a consistent convergence. Furthermore, an inner-outer border exchange scheme is performed to obviate the border effect between neighboring GPUs. The proposed FL-MISR is applied to the computed tomography (CT) system by super-resolving the projections acquired by subpixel detector shift. The SR reconstruction is performed on the fly during the CT acquisition such that no additional computation time is introduced. FL-MISR is extensively evaluated from different aspects and experimental results demonstrate that FL-MISR effectively improves the spatial resolution of CT systems in modulation transfer function (MTF) and visual perception. Comparing to a multi-core CPU implementation, FL-MISR achieves a more than 50x speedup on an off-the-shelf 4-GPU system.

arXiv:2108.03551공식 코드 · Papers with Code

Disentangled High Quality Salient Object Detection

Aiming at discovering and locating most distinctive objects from visual scenes, salient object detection (SOD) plays an essential role in various computer vision systems. Coming to the era of high resolution, SOD methods are facing new challenges. The major limitation of previous methods is that they try to identify the salient regions and estimate the accurate objects boundaries simultaneously with a single regression task at low-resolution. This practice ignores the inherent difference between the two difficult problems, resulting in poor detection quality. In this paper, we propose a novel deep learning framework for high-resolution SOD task, which disentangles the task into a low-resolution saliency classification network (LRSCN) and a high-resolution refinement network (HRRN). As a pixel-wise classification task, LRSCN is designed to capture sufficient semantics at low-resolution to identify the definite salient, background and uncertain image regions. HRRN is a regression task, which aims at accurately refining the saliency value of pixels in the uncertain region to preserve a clear object boundary at high-resolution with limited GPU memory. It is worth noting that by introducing uncertainty into the training process, our HRRN can well address the high-resolution refinement task without using any high-resolution training data. Extensive experiments on high-resolution saliency datasets as well as some widely used saliency benchmarks show that the proposed method achieves superior performance compared to the state-of-the-art methods.

arXiv:2108.03786

Multi-Slice Net: A novel light weight framework for COVID-19 Diagnosis

This paper presents a novel lightweight COVID-19 diagnosis framework using CT scans. Our system utilises a novel two-stage approach to generate robust and efficient diagnoses across heterogeneous patient level inputs. We use a powerful backbone network as a feature extractor to capture discriminative slice-level features. These features are aggregated by a lightweight network to obtain a patient level diagnosis. The aggregation network is carefully designed to have a small number of trainable parameters while also possessing sufficient capacity to generalise to diverse variations within different CT volumes and to adapt to noise introduced during the data acquisition. We achieve a significant performance increase over the baselines when benchmarked on the SPGC COVID-19 Radiomics Dataset, despite having only 2.5 million trainable parameters and requiring only 0.623 seconds on average to process a single patient's CT volume using an Nvidia-GeForce RTX 2080 GPU.

arXiv:2108.03434

NASOA: Towards Faster Task-oriented Online Fine-tuning with a Zoo of Models

Fine-tuning from pre-trained ImageNet models has been a simple, effective, and popular approach for various computer vision tasks. The common practice of fine-tuning is to adopt a default hyperparameter setting with a fixed pre-trained model, while both of them are not optimized for specific tasks and time constraints. Moreover, in cloud computing or GPU clusters where the tasks arrive sequentially in a stream, faster online fine-tuning is a more desired and realistic strategy for saving money, energy consumption, and CO2 emission. In this paper, we propose a joint Neural Architecture Search and Online Adaption framework named NASOA towards a faster task-oriented fine-tuning upon the request of users. Specifically, NASOA first adopts an offline NAS to identify a group of training-efficient networks to form a pretrained model zoo. We propose a novel joint block and macro-level search space to enable a flexible and efficient search. Then, by estimating fine-tuning performance via an adaptive model by accumulating experience from the past tasks, an online schedule generator is proposed to pick up the most suitable model and generate a personalized training regime with respect to each desired task in a one-shot fashion. The resulting model zoo is more training efficient than SOTA models, e.g. 6x faster than RegNetY-16GF, and 1.7x faster than EfficientNetB3. Experiments on multiple datasets also show that NASOA achieves much better fine-tuning results, i.e. improving around 2.1% accuracy than the best performance in RegNet series under various constraints and tasks; 40x faster compared to the BOHB.

arXiv:5508.33299

Triton: An Intermediate Language and Compiler for Tiled Neural Network Computations

The validation and deployment of novel research ideas in the field of Deep Learning is often limited by the availability of efficient compute kernels for certain basic primitives. In particular, operations that cannot leverage existing vendor libraries (e.g., cuBLAS, cuDNN) are at risk of facing poor device utilization unless custom implementations are written by experts – usually at the expense of portability. For this reason, the development of new programming abstractions for specifying custom Deep Learning workloads at a minimal performance cost has become crucial. We present Triton, a language and compiler centered around the concept of tile, i.e., statically shaped multi-dimensional sub-arrays. Our approach revolves around (1) a C-based language and an LLVM-based intermediate representation (IR) for expressing tensor programs in terms of operations on parametric tile variables and (2) a set of novel tile-level optimization passes for compiling these programs into efficient GPU code. We demonstrate how Triton can be used to build portable implementations of matrix multiplication and convolution kernels on par with hand-tuned vendor libraries (cuBLAS / cuDNN), or for efficiently implementing recent research ideas such as shift convolutions.

arXiv:2108.03856

BenchENAS: A Benchmarking Platform for Evolutionary Neural Architecture Search

Neural architecture search (NAS), which automatically designs the architectures of deep neural networks, has achieved breakthrough success over many applications in the past few years. Among different classes of NAS methods, evolutionary computation based NAS (ENAS) methods have recently gained much attention. Unfortunately, the issues of fair comparisons and efficient evaluations have hindered the development of ENAS. The current benchmark architecture datasets designed for fair comparisons only provide the datasets, not the ENAS algorithms or the platform to run the algorithms. The existing efficient evaluation methods are either not suitable for the population-based ENAS algorithm or are too complex to use. This paper develops a platform named BenchENAS to address these issues. BenchENAS aims to achieve fair comparisons by running different algorithms in the same environment and with the same settings. To achieve efficient evaluation in a common lab environment, BenchENAS designs a parallel component and a cache component with high maintainability. Furthermore, BenchENAS is easy to install and highly configurable and modular, which brings benefits in good usability and easy extensibility. The paper conducts efficient comparison experiments on eight ENAS algorithms with high GPU utilization on this platform. The experiments validate that the fair comparison issue does exist, and BenchENAS can alleviate this issue. A website has been built to promote BenchENAS at https://benchenas.com, where interested researchers can obtain the source code and document of BenchENAS for free.

arXiv:2108.03645

Online Evolutionary Batch Size Orchestration for Scheduling Deep Learning Workloads in GPU Clusters

Efficient GPU resource scheduling is essential to maximize resource utilization and save training costs for the increasing amount of deep learning workloads in shared GPU clusters. Existing GPU schedulers largely rely on static policies to leverage the performance characteristics of deep learning jobs. However, they can hardly reach optimal efficiency due to the lack of elasticity. To address the problem, we propose ONES, an ONline Evolutionary Scheduler for elastic batch size orchestration. ONES automatically manages the elasticity of each job based on the training batch size, so as to maximize GPU utilization and improve scheduling efficiency. It determines the batch size for each job through an online evolutionary search that can continuously optimize the scheduling decisions. We evaluate the effectiveness of ONES with 64 GPUs on TACC's Longhorn supercomputers. The results show that ONES can outperform the prior deep learning schedulers with a significantly shorter average job completion time.

arXiv:2108.02191

Random Offset Block Embedding Array (ROBE) for CriteoTB Benchmark MLPerf DLRM Model : 1000$\times$ Compression and 3.1$\times$ Faster Inference

Deep learning for recommendation data is one of the most pervasive and challenging AI workload in recent times. State-of-the-art recommendation models are one of the largest models matching the likes of GPT-3 and Switch Transformer. Challenges in deep learning recommendation models (DLRM) stem from learning dense embeddings for each of the categorical tokens. These embedding tables in industrial scale models can be as large as hundreds of terabytes. Such large models lead to a plethora of engineering challenges, not to mention prohibitive communication overheads, and slower training and inference times. Of these, slower inference time directly impacts user experience. Model compression for DLRM is gaining traction and the community has recently shown impressive compression results. In this paper, we present Random Offset Block Embedding Array (ROBE) as a low memory alternative to embedding tables which provide orders of magnitude reduction in memory usage while maintaining accuracy and boosting execution speed. ROBE is a simple fundamental approach in improving both cache performance and the variance of randomized hashing, which could be of independent interest in itself. We demonstrate that we can successfully train DLRM models with same accuracy while using $1000 \times$ less memory. A $1000\times$ compressed model directly results in faster inference without any engineering effort. In particular, we show that we can train DLRM model using ROBE array of size 100MB on a single GPU to achieve AUC of 0.8025 or higher as required by official MLPerf CriteoTB benchmark DLRM model of 100GB while achieving about $3.1\times$ (209\%) improvement in inference throughput.

arXiv:2108.02037

The MIT Supercloud Dataset

Artificial intelligence (AI) and Machine learning (ML) workloads are an increasingly larger share of the compute workloads in traditional High-Performance Computing (HPC) centers and commercial cloud systems. This has led to changes in deployment approaches of HPC clusters and the commercial cloud, as well as a new focus on approaches to optimized resource usage, allocations and deployment of new AI frame- works, and capabilities such as Jupyter notebooks to enable rapid prototyping and deployment. With these changes, there is a need to better understand cluster/datacenter operations with the goal of developing improved scheduling policies, identifying inefficiencies in resource utilization, energy/power consumption, failure prediction, and identifying policy violations. In this paper we introduce the MIT Supercloud Dataset which aims to foster innovative AI/ML approaches to the analysis of large scale HPC and datacenter/cloud operations. We provide detailed monitoring logs from the MIT Supercloud system, which include CPU and GPU usage by jobs, memory usage, file system logs, and physical monitoring data. This paper discusses the details of the dataset, collection methodology, data availability, and discusses potential challenge problems being developed using this data. Datasets and future challenge announcements will be available via https://dcc.mit.edu.

arXiv:2108.01941공식 코드 · Papers with Code

Automatic cerebral hemisphere segmentation in rat MRI with lesions via attention-based convolutional neural networks

We present MedicDeepLabv3+, a convolutional neural network that is the first completely automatic method to segment cerebral hemispheres in magnetic resonance (MR) volumes of rats with lesions. MedicDeepLabv3+ improves the state-of-the-art DeepLabv3+ with an advanced decoder, incorporating spatial attention layers and additional skip connections that, as we show in our experiments, lead to more precise segmentations. MedicDeepLabv3+ requires no MR image preprocessing, such as bias-field correction or registration to a template, produces segmentations in less than a second, and its GPU memory requirements can be adjusted based on the available resources. We optimized MedicDeepLabv3+ and six other state-of-the-art convolutional neural networks (DeepLabv3+, UNet, HighRes3DNet, V-Net, VoxResNet, Demon) on a heterogeneous training set comprised by MR volumes from 11 cohorts acquired at different lesion stages. Then, we evaluated the trained models and two approaches specifically designed for rodent MRI skull stripping (RATS and RBET) on a large dataset of 655 MR rat brain volumes. In our experiments, MedicDeepLabv3+ outperformed the other methods, yielding an average Dice coefficient of 0.952 and 0.944 in the brain and contralateral hemisphere regions. Additionally, we show that despite limiting the GPU memory and the training data, our MedicDeepLabv3+ also provided satisfactory segmentations. In conclusion, our method, publicly available at https://github.com/jmlipman/MedicDeepLabv3Plus, yielded excellent results in multiple scenarios, demonstrating its capability to reduce human workload in rat neuroimaging studies.